«INFLAMMATORY PROTEINS, GENETIC VARIATION, AND ENVIRONMENTAL INFLUENCES ON HEALTH CARE ASSOCIATED INFECTION DEVELOPMENT IN SEPSIS A Dissertation ...»
Note: Data are reported as mean ± standard deviation, median (interquartile range), or count (percentage).
than twice compared to those not developing HAI, and required a longer duration of mechanical ventilation. There was a higher rate of steroid use in those developing HAI compared to those who did not (82.4% vs. 42.6%, p = 0.005). Patients developing HAI also had a trend (p = 0.06) towards higher ICU readmissions rate within 28 days as compared to those who did not develop HAI.
Differences in Variables by Pro- and Anti-inflammatory Cytokine Quartiles among Those Who Did and Did Not Develop HAI A summary is provided in Table 4-7 comparing participants with high baseline systemic inflammation (fourth quartile) to participants without high baseline systemic inflammation (first-third quartiles) for both the pro-inflammatory cytokine IL-6 and the anti-inflammatory cytokine IL-10. Steroid use was the only significant difference among participants with a high anti-inflammatory response versus those without (84.2% vs.
40.7%, p = 0.0013). Among those with a high pro-inflammatory response there were several significant differences in comparison to those with a lower pro-inflammatory response. Organ dysfunction was higher, and there was a three fold higher number of participants requiring vasopressors (73.7% vs. 23.7%, p 0.0001).There was also a higher percentage of steroid use (73.7% vs. 44.1%, p = 0.03), as well as participants with arrythmias (52.6% vs. 18.6%, p = 0.007). ICU Survival was similar for all participants regardless of baseline pro- or anti-inflammatory cytokine level.
Cytokine and Genotype Measurements
Cytokines were measured by Luminex in batches. Details are described in the methods section. The detection limits for each cytokine was 3.2 pg/ml and 10,000 pg/ml.
A total of 5 (6.4%) of IL-6 levels and 29 (37.2%) of IL-10 levels were below the detection limit. In these cases, a surrogate of 3.2 pg/ml was used for statistical purposes.
DNA isolation occurred in batches during recruitment and after recruitment was complete. Genotyping was perfomed after study recruitment was complete using End Point Genotyping by Real-Time PCR. The LightCycler® 480 software (version LCS480 18.104.22.168) by Roche (Mannheim, Germany) provided automated genotyping calls for each participant. A call of either Allele X, Allele Y, Both Alleles, Unknown, or Negative was provided for each of the 96 wells. Gentotypes for Allele X, Allele Y, and Both Alleles were described in Table 3.2. A negative call was received for the negative controls, empty wells, and for other possible reasons such as sample quality, inhibition, or primer/dimer formation. Unknown calls were received when the software was unable to determine the genotype.
A test run was performed to optimize the PRC reaction which included the first 10 participant’s samples under conditions described in the methods section. Replication was not observed until 30 cycles during the test run; therefore, the cycle time was increased to 50 from 45 cycles based on expert guidance from a MRC scientists who assisted with Table 4-7. Differences in Variables Pro- and Anti-inflammatory Cytokine Quartiles.
* Duration of mechanical ventilation includes data for 34 subjects who received mechanical ventilation, n=24 versus 10, respectively.
LightCycler initial set-up and with calculations required for initial reaction volumes. The genotype results for IL-6 under these conditions included 32 Negatives and 5 Unknowns.
Applied Biosystems technical support advised that although replication did not begin until 30 cycles, it is more appropriate for end point genotyping not to increase cycle time beyond 45 cycles. All samples were re-tested for IL-6 genotyping using the original test conditions (45 cycles) and resulted in only 2 undetermined genotypes (1 negative and 1 unknown call). These 2 missing genotypes were available from the initial test run.
Genotyping for IL-10 was performed as described in the methods section. The initial undetermined calls were 30 unknown and 1 negative. The reaction was optimized by increasing the temperature to 62°C, reducing the number of unknowns to 7. Manual calls were required for these 7 subjects: 2, 22, 27, 31, 50, 58, and 69. These subjects were manually called heterozygotes based upon visual clustering and endpoint fluorescence values. All unknown calls had a confidence score less than 0.50 and endpoint fluorescence values were not significantly higher than each other. Although the software algorithm is proprietary, genotypes that were automatically called included an endpoint fluorescence of at least one or more fold higher than the lower endpoint fluorescence value. The endpoint fluorescence value for those called as heterozygous contained differences but they were generally much less than a fold difference.
Genotype Allele Frequencies
The allele frequencies are shown for rs1800795 and rs1800896 genotypes in Table 4-8 and 4-9, respectively. These tables also show allele frequencies for all subjects as well as white and black sepsis patients who do and do not develop HAI. Note that there were no significant differences between the percentage of blacks and white sepsis patients who do and do not develop HAI (23.3% vs. 20.8%, p = 0.79).
For rs1800795, there were no significant allele frequency differences among those who do and do not develop HAI (p = 0.59); however, when examining racial differences in genotype and controlling for those who do and do not develop HAI differences were noted. There was a significant difference in rs1800795 genotype among black patients with sepsis who did not develop HIA compared to whites patients with sepsis who did not develop HAI (p = 0.0056). Specifically, black patients had a lower CG (17.4% vs.
42.1%) and higher GG (82.6% vs. 42.1%) than white patients. There were no racial differences when comparing white and black sepsis patients who developed HAI (p = 1.0).
For rs1800896, there were no significant allele frequency differences among those who do and do not develop HAI (p = 0.16). There were no significant racial differences among those who did (p = 1.0) and did not develop HAI (p = 0.41).
Table 4-8. RS1800795 Genotype and Allele Frequencies.
Genotype Comparisons to HapMap 3 Reference Population There were differences in the genotype of sepsis patients as compared to a normal HapMap3 reference population (see Table 4-10). White (and not black) patients with sepsis had a significantly different (p = 0.02) IL-6 genotype with higher GG (41.7% versus 24.8%) and lower CC (14.6% versus 31.9%) genotypes when compared to a normal reference. When examining IL-10 genotypes, black (and not white) patients with sepsis had a significantly higher GG (30.0% versus 12.2%) and lower AA (13.3% versus 44.9%) genotype when compared to a normal reference.
Baseline Cytokine Levels by Genotype and Haplotypes
Plasma cytokine levels were right skewed and required log transformation for statistcs requiring a normal disturbution. Table 4-11 provides a summary of plasma IL-6 and IL-10 levels. Median IL-6 levels were higher than median IL-10 levels. Table 4-12 provides cytokine levels and their ratio for each genotype. Median IL-6 levels were highest among participants with the CC IL-6 genotype and also among participants with the GG IL-10 genotype. Median IL-10 levels were highest among participants with the AA IL-10 genotype and also among participants with the CG IL-6 genotype. Figures 4-3, 4-4, 4-5 and 4-6 provide box-plots of pro- and anti-inflammatory cytokines and their ratios by for IL-6 and IL-10 genotypes. Each figure contains two images. The top image shows the skewed data distribution prior to log transformation, and the lower image shows the log transformed data distribution.
The primary goal of aim one was to investigate whether baseline protein expression levels of pro-inflammatory cytokines, anti-inflammatory cytokines, or their ratios influence the development of subsequent HAI in patients with sepsis.
There was no significant difference in levels of pro-inflammatory cytokine, antiinflammatory cytokine, or their ratio among subjects who did and did not develop at least one HAI during their ICU stay. This aim was explored by comparing lower quartiles to the higher fourth quartile for proinflammatory cytokine IL-6 and anti-inflammatory cytokine IL-10, as well as comparing their ratios. Patients in the fourth quartile were considered to have an exaggerated inflammatory response as compared to those in other quartiles. Specifically, an exaggerated pro-inflammatory response was present in 6 (31.6%) compared to 11 (18.6%) participants without an exaggerated pro-inflammatory response who developed subsequent HAI. This difference was not significant (p = 0.55).
Likewise, an exaggerated anti-inflammatory response was present in 5 (26.3%) compared to 12 (20.3) participants without an exaggerated anti-inflammatory response who developed subsequent HAI. This difference was also not significant (p = 0.43). There was also no significant difference in the log of proinflammatory to anti-inflammatory Table 4-10. Genotype Comparisons of IL-6 and IL-10 SNPs among Sepsis and HapMap Reference Population.
Note: The reference population is the HapMap 3. The HapMap 3 sample included in this comparison includes normal individuals from African ancestry in Southwest USA (black) and Utah residents with Northern and Western European ancestry (white).109 Table 4-11. Range of Plasma IL6 and IL10 Levels.
Figure 4-6. Box Plot of Plasma IL-6:IL10 ratios by IL-10 Genotypes Pre and Post Log Transformation.
ratios among those who did and did not develop subsequent HAI (1.8 ± 1.5 vs. 1.6 ± 1.8, p = 0.55). Cytokine levels were right skewed; thus, non-parametric tests were performed to provide a comparison of median values among participants who did and did not develop HAI. Median cytokine measurements are shown in Table 4-13. There were no significant differences in baseline pro-inflammatory cytokines, anti-inflammatory cytokines, or their ratios among participants who did and did not develop at least one HAI during their ICU stay.
The goal of aim two was to investigate the variance in cytokine genes to determine if they influence levels of protein expression or development of HAI.
The variance in cytokine genes were determined by SNP analysis. The distribution of subjects per genotype is shown in Table 4-14. An ANOVA was performed to examine the difference in cytokine means for each genotype. Table 4-15 and Figure 4-7 summarize the cytokine levels for each SNP. There were no significant differences in plasma IL-6 levels based on SNP rs1800795. Both homozygous AA and GG genotypes for IL-10 SNP rs1800896 were significantly higher (0.02) that the heterozygous GA.
Table 4-14 summarizes development of HAI by genotype and haplotype. There were no statistically significant differences among genotypes or haplotypes for development of HAI. There were no CC_AA, CC_GA, or GG_AA haplotypes among the 17 participants who developed HAI. The most common haplotype, was the heterozygous CG_GA. This haplotype is also where the highest percentage of HAIs occurred but this was not significant.
The goal of aim three was to investigate the effects of protein expression levels, genetic variation, and environment on development of HAI. A series of Cox regression analyses were performed among those who did and did not develop HAIs during ICU stay (or up to 28 days in those with a prolonged ICU stay) controlling for a number of potentially confounding variables. Table 4-16 provides a summary of variables testing for inclusion into the multivariate model. Only variables with a p-value of less than 0.25 were included in the multivariate regression model (Table 4-17). There were only four females in the study and it was not appropriate to include gender in the model (HR = 1313884).
Aim three included a series of questions pertaining to risk of developing HAIs. In general, a hazard ratio greater than 1 indicates a higher risk of developing an event, in this case health care associated infections, and a hazard ratio less than 1 indicates a lower Table 4-13. Comparison of Cytokine Levels among Subjects Developing HAI.
Note: Median and interquartile range shown.
Table 4-14. SNP Genotypes and Haplotypes for All Subjects by HAI Development and Pro- and Anti-inflammatory Cytokine Quartiles.
Note: IL-6 Genotypes are compared to GG genotype.
risk of developing an event. Each question was answered based on univariate cox
• The hazard ratio to predict development of HAIs for APACHE II is 1.069. For each 10 point increase in APACHE II score the risk ratio or hazard ratio is e0.06656 x 10 = 1.956. For a 10 point increased in APACHE II score, there is a 95.6 percent increase in the risk of developing a HAI.
• The hazard ratio to predict development of HAI for each additional invasive devise is e-0.03742 = 0.963. For an additional invasive devise score there was a 3.7% lower risk of developing a HAI. The cumulative invasive devise score at the time of HAI had been compared to the invasive devise score at ICU discharge among participants who did not develop a HAI.
• The hazard ratio to predict development of HAIs given IL-6 -174G genotype CG compared to GG is e1.43513 = 4.200, and for IL6 genotype CC compared to GG the hazard ratio is e1.56334 = 4.775. Presence of the GG genotype has a four-fold increase in risk for development of HAI.
• The hazard ratio to predict development of HAIs given IL-10 -1082G genotypeIL-10 GA compared to GG is e1.39081 = 4.018, AA compared to GG is e1.34786 = 3.849. Presence of the GG genotype has a four-fold increase in risk for development of HAI The hazard ratio to predict development of HAIs for each 10 point increase in pro-inflammatory cytokine based on plasma IL-6 is e0.0001308 x 10 = 1.001. For a 10 point increase in IL-6 there is a 0.1% increase in developing a HAI.